The Experts below are selected from a list of 20094 Experts worldwide ranked by ideXlab platform
Moshe Sipper - One of the best experts on this subject based on the ideXlab platform.
-
a Single Iteration threshold hamming network
IEEE Transactions on Neural Networks, 1995Co-Authors: Isaac Meilijson, Eytan Ruppin, Moshe SipperAbstract:We analyze in detail the performance of a Hamming network classifying inputs that are distorted versions of one of its m stored memory patterns, each being a binary vector of length n. It is shown that the activation function of the memory neurons in the original Hamming network may be replaced by a simple threshold function. By judiciously determining the threshold value, the "winner-take-all" subnet of the Hamming network (known to be the essential factor determining the time complexity of the network's computation) may be altogether discarded. For m growing exponentially in n, the resulting threshold Hamming network correctly classifies the input pattern in a Single Iteration, with probability approaching 1. >
-
Single Iteration threshold hamming networks
Neural Information Processing Systems, 1992Co-Authors: Isaac Meilijson, Eytan Ruppin, Moshe SipperAbstract:We analyze in detail the performance of a Hamming network classifying inputs that are distorted versions of one of its m stored memory patterns. The activation function of the memory neurons in the original Hamming network is replaced by a simple threshold function. The resulting Threshold Hamming Network (THN) correctly classifies the input pattern, with probability approaching 1, using only O(m ln m) connections, in a Single Iteration. The THN drastically reduces the time and space complexity of Hamming Network classifiers.
Sugih Jamin - One of the best experts on this subject based on the ideXlab platform.
-
a new scheduling algorithm for input queued switches with mixed unicast and multicast traffic
High Performance Switching and Routing, 2018Co-Authors: Jie Xiao, Kwan L Yeung, Sugih JaminAbstract:We consider an $N\times N$ input-queued switch with $N$ dedicated unicast virtual output queues (VOQs) and one shared multicast queue (MQ) at each input port. An efficient two-bit Single-Iteration (2BSI) scheduling algorithm is proposed to concurrently schedule both unicast and multicast traffic. In the request phase, a two-bit request message is used to indicate not only the type of the request (unicast/multicast) but also its importance (strong/weak). In the grant phase, multicast request is granted first, then strong unicast request, and finally weak unicast request. To minimize inconsistencies in the distributed arbitration process, the notion of preferred unicast/multicast relationship is adopted to desynchronize/synchronize the arbitration decisions made by different inputs/outputs. As compared to the existing schedulers, our 2BSI is one of the simplest algorithms to implement, and yet extensive simulation results show that it provides one of the best delay-throughput performances.
-
scheduling mixed unicast and multicast traffic with variable size packets in input queued switches
High Performance Switching and Routing, 2018Co-Authors: Jie Xiao, Kwan L Yeung, Sugih JaminAbstract:We consider scheduling mixed unicast and multicast traffic with variable-size packets in an input-queued switch. When variable-size packets arrive at a switch input port, they will be segmented into cells (fixed-size packets), sent across the switch fabric, and reassembled at outputs. A scheduling algorithm should focus on optimizing packet performance rather than cell performance. In this paper, packet-mode scheduling is adopted such that cells of the same packet are sent back-to-back in consecutive slots. For efficiency, an iterative scheduling algorithm called three-bit Single-Iteration (3BSI) is proposed to concurrently schedule both unicast and multicast traffic. To the best of our knowledge, 3BSI is the first packet-mode scheduling algorithm for handling mixed traffic with variable-size packets. Despite its simplicity, extensive simulation shows that 3BSI provides excellent delay-throughput performance.
-
highest rank first a new class of Single Iteration scheduling algorithms for input queued switches
IEEE Access, 2018Co-Authors: Fujie Fan, Kwan L Yeung, Sugih JaminAbstract:In this paper, we study a new class of Single-Iteration scheduling algorithms for input-queued switches based on a new arbitration idea called highest rank first (HRF). We first demonstrate the effectiveness of HRF by a simple algorithm named Basic-HRF. In Basic-HRF, virtual output queues (VOQs) at an input port are ranked according to their queue sizes. The rank of a VOQ, coded by $\log (N+1)$ bits, where $N$ is the switch size, is sent to the corresponding output as a request. Unlike all existing iterative algorithms, the winner is selected based on the ranks of the requests/grants. We show that the rank-based arbitration outperforms the widely adopted queue-based arbitration. To improve the performance under heavy load and maximize the match size, Basic-HRF is integrated with an embedded round-robin scheduler. The resulting HRF algorithm is shown to beat almost all existing Single-Iteration algorithms. But, the complexity of HRF is high due to the use of multi-bit requests. A novel request encoding/decoding mechanism is then designed to reduce the request size to a Single bit while keeping the original performance of HRF. A unique feature of the resulting coded HRF (CHRF) algorithm is that the Single-bit request indicates an increase or decrease of a VOQ rank, rather than an empty VOQ or not. We show that the CHRF is the most efficient Single-bit-Single-Iteration algorithm.
-
pipelined scheduler for unicast and multicast traffic in input queued switches
Global Communications Conference, 2016Co-Authors: Jie Xiao, Kwan L Yeung, Sugih JaminAbstract:We focus on designing efficient integrated schedulers for handling mixed unicast and multicast traffic. We consider an input-queued switch with a multicast-capable switch fabric. At each input port of the switch, there are N dedicated unicast VOQs and one shared multicast queue (MQ). An existing approach to the design of integrated scheduler (i.e., a sequential scheduler) is to run two component schedulers, one for multicast and one for unicast, sequentially in each time slot. To minimize the head-of-line blocking of multicast traffic, the multicast scheduler always runs first. But sequentially running two schedulers in each time slot is challenging, especially when the slot duration is small. In this paper, we first propose a pipelined integration of the two component schedulers (i.e., a pipelined scheduler), which allows twice the amount of time for each scheduler to execute. We then extend an existing Single-bit- Single-Iteration unicast scheduler to ensure that even in the presence of multicast traffic, unicast traffic will be starvation-free. This is achieved by giving unicast traffic priority over multicast periodically. Finally, we present arguably the first Single-bit-Single-Iteration multicast scheduling algorithm. Extensive simulation results show that our pipelined scheduler is efficient and provides delay-throughput performance comparable to the sequential scheduler.
Isaac Meilijson - One of the best experts on this subject based on the ideXlab platform.
-
a Single Iteration threshold hamming network
IEEE Transactions on Neural Networks, 1995Co-Authors: Isaac Meilijson, Eytan Ruppin, Moshe SipperAbstract:We analyze in detail the performance of a Hamming network classifying inputs that are distorted versions of one of its m stored memory patterns, each being a binary vector of length n. It is shown that the activation function of the memory neurons in the original Hamming network may be replaced by a simple threshold function. By judiciously determining the threshold value, the "winner-take-all" subnet of the Hamming network (known to be the essential factor determining the time complexity of the network's computation) may be altogether discarded. For m growing exponentially in n, the resulting threshold Hamming network correctly classifies the input pattern in a Single Iteration, with probability approaching 1. >
-
Single Iteration threshold hamming networks
Neural Information Processing Systems, 1992Co-Authors: Isaac Meilijson, Eytan Ruppin, Moshe SipperAbstract:We analyze in detail the performance of a Hamming network classifying inputs that are distorted versions of one of its m stored memory patterns. The activation function of the memory neurons in the original Hamming network is replaced by a simple threshold function. The resulting Threshold Hamming Network (THN) correctly classifies the input pattern, with probability approaching 1, using only O(m ln m) connections, in a Single Iteration. The THN drastically reduces the time and space complexity of Hamming Network classifiers.
Eytan Ruppin - One of the best experts on this subject based on the ideXlab platform.
-
a Single Iteration threshold hamming network
IEEE Transactions on Neural Networks, 1995Co-Authors: Isaac Meilijson, Eytan Ruppin, Moshe SipperAbstract:We analyze in detail the performance of a Hamming network classifying inputs that are distorted versions of one of its m stored memory patterns, each being a binary vector of length n. It is shown that the activation function of the memory neurons in the original Hamming network may be replaced by a simple threshold function. By judiciously determining the threshold value, the "winner-take-all" subnet of the Hamming network (known to be the essential factor determining the time complexity of the network's computation) may be altogether discarded. For m growing exponentially in n, the resulting threshold Hamming network correctly classifies the input pattern in a Single Iteration, with probability approaching 1. >
-
Single Iteration threshold hamming networks
Neural Information Processing Systems, 1992Co-Authors: Isaac Meilijson, Eytan Ruppin, Moshe SipperAbstract:We analyze in detail the performance of a Hamming network classifying inputs that are distorted versions of one of its m stored memory patterns. The activation function of the memory neurons in the original Hamming network is replaced by a simple threshold function. The resulting Threshold Hamming Network (THN) correctly classifies the input pattern, with probability approaching 1, using only O(m ln m) connections, in a Single Iteration. The THN drastically reduces the time and space complexity of Hamming Network classifiers.
Zhaoyang Zhang - One of the best experts on this subject based on the ideXlab platform.
-
an efficient Single Iteration Single bit request scheduling algorithm for input queued switches
Journal of Network and Computer Applications, 2013Co-Authors: K L Yeung, Zhaoyang ZhangAbstract:Aiming at minimizing communication overhead of iterative scheduling algorithms for input-queued packet switches, an efficient Single-Iteration Single-bit request scheduling algorithm called Highest Rank First with Request Compression 1 (HRF/RC1) is proposed. In HRF/RC1, scheduling priority is given to the preferred input-output pair first, where each input has a distinct preferred output in each time slot. If an input does not have backlogged packets for its preferred output, each of its non-empty VOQs sends a Single-bit request to the corresponding output. This Single bit distinguishes one longest VOQ from other non-empty VOQs among an input port. If an output receives a request from its preferred input, it grants this input. Otherwise, it gives the higher priority to the longest VOQ than other non-empty VOQs. Similarly, an input accepts the grant following the same propriety sequence. In case of a tie, the winner is selected randomly. Compared with other Single-Iteration algorithms with comparable communication overhead, we show by simulations that HRF/RC1 always gives the best delay-throughput performance.